Sugestii pentru un model de adnotare a emoțiilor și sentimentelor în romanul românesc de consum din lungul secol al XIX-lea
Proposals for a framework for annotating emotions and sentiments in Romanian popular novels of the long nineteenth century

Summary/Abstract: Given the significant role of emotions and sentiments in Romanian popular novels from the long 19th century, sociological literary studies stand to benefit greatly from the automatic extraction of affective features in texts, a task made possible through Sentiment Analysis (SA) methods. In the absence of a SA model tailored to Romanian literature, this paper proposes a set of guidelines for developing such a model, drawing on both the state-of-the-art in computational literary studies and the specific characteristics of Romanian popular fiction. We argue that the most appropriate theoretical framework for this purpose is the Hourglass Model Revisited, which offers a robust foundation for manual annotation and the subsequent use of annotated texts in machine learning applications. Recognizing that the quality of manual annotation strongly influences the effectiveness of automatic feature extraction, we place particular emphasis on the creation of rigorous annotation guidelines, supplemented by numerous examples from three sentimental novels by N. G. Rădulescu-Niger and detailed discussion aimed at increasing analytical precision. The overarching goal of this research is to gain deeper insights into the ideas, values, and worldview of 19th-century Romanian readers by examining their expected emotional responses to popular fiction.

• Details
• Contents